Second Generation AMD Bulldozer – Vishera

[teletran8]

I must admit after all this time its not looking good. With this kind of delay, with this kind of product for whatever reason, someone somewhere has f**ked up big time.

Either the product doesnt work. Its not fast enough. The Fabs are screwed or there is a major flaw in it. It will all come out in the end.

I dont think its due to a spelling mistake on the retail boxes.

Everyone would like to know the reason, but then again it's obviously a negative reason...yet no one (GF/AMD) is saying are they. Both of their buisnesses / corporations future is determined by product. Time will tell obviously. Only if yields are so bad they cannot make money AMD would be in trouble. Or simply AMD and/or GF has the time and is not into rushing, it could be as simple as that tho.

My opinion is that the new bulldozer "module" design leads to lower yields per wafer, or simply GF is slacking BIG time. They've obviously been able to yield a few though so things look fine to me...hehe.

 

[daglesj]

Buying Intel and/or waiting on Jesus is more of a travesty than waiting for AMD's next cutting edge CPU design imho.

I've been a big AMD fan (still am) and buyer for many many years. But my friend I think you are going to be a bit dissapointed here. Not that it wont be the first time but.........

The lead up to this draws two conclusions. Its fantastic and been worth the wait and exceeds expectations.

Or

Its a total flop and its another 12 months before we get a true taste of what it could have been, by which time it will be a bit late.

Past experience leads me to think the second is probably the one to put money on. I truly hope I'm wrong but I'm not getting my hopes up.

 

[drescherjm]

I still believe the 8150 will be somewhere between those two.. Competitive at less than 5 threads because of higher turbo. In 6 or more threads it should do well.

 

[daglesj]

Well here's hoping. It would be a fun time ahead if it does work out well in the end.

If it doesnt it isnt the end of the world.

 

[teletran8]

I've been a big AMD fan (still am) and buyer for many many years. But my friend I think you are going to be a bit dissapointed here. Not that it wont be the first time but.........

The lead up to this draws two conclusions. Its fantastic and been worth the wait and exceeds expectations.

Or

Its a total flop and its another 12 months before we get a true taste of what it could have been, by which time it will be a bit late.

Past experience leads me to think the second is probably the one to put money on. I truly hope I'm wrong but I'm not getting my hopes up.

People thought the world was flat at one time, so I don't put much stock into the past.

This is quite hilarious to me, the Intel corporate brainwashing here is out of control. I'll be the sanity since most* (you) people posting here have lost your minds to the big blue machine.

I'll be coming from an X3 @ 4.00GHz and going to a BD 8 Core that does 4.5 - 5.0 GHz on air and who knows what on water imo.

If you don't understand that it will rock, well then that's your problem.

And if all you have is made up expectations from incestuous Intel fans, and spouting off your negativity on AMD [H] forums about a CPU you haven't personally benched / used you just might be a fool.

AMD demo'd the chip OC'd it to 8.4+ if you can't believe reality, that the big red machine stomped every CPU in existence in OC well there's no helping you.

 

[mrbigshot]

I'll be coming from an X3 @ 4.00GHz and going to a BD 8 Core that does 4.5 - 5.0 GHz on air and who knows what on water imo.

If you don't understand that it will rock, well then that's your problem.

AMD demo'd the chip OC'd it to 8.4+ if you can't believe reality, that the big red machine stomped every CPU in existence in OC well there's no helping you.

at the same demo they also were only able to get 4.8 under water.

 

[mzs_biteme]

....AMD demo'd the chip OC'd it to 8.4+ if you can't believe reality, that the big red machine stomped every CPU in existence in OC well there's no helping you.

Didn't they have to turn off all but two cores for get to that overly useless OC???
And if you think the ability to OC to ridicules speeds has anything to do actually performance, than the "big RED brainwashing" definitely worked on you
...

 

[drescherjm]

Didn't they have to turn off all but two cores for get to that overly useless OC???

They turned off all but 2 cores used very unrealistic cooling and did not even load the 2 cores. To me it was a completely useless demonstration as far as anything real world. I believe it was a good publicity stunt to raise the spirits of the fanbase.

 

[teletran8]

at the same demo they also were only able to get 4.8 under water.

That's funny because AMD states: "We also achieved clock frequencies well above 5GHz using only air or sub-$100 water cooling solutions."
From the AMD BLOG

A cute troll says 4.8 on water...sounds more like a Sandy Bridge chip your (running) thinking of

 

[drescherjm]

AMD demo'd the chip OC'd it to 8.4+ if you can't believe reality, that the big red machine stomped every CPU in existence in OC well there's no helping you.

Wasn't the previous record a Pentuim 4 Celeron at 8.2GHz. I would not call 8.4Ghz a stomp over that.

 

[mzs_biteme]

....A cute troll says 4.8 on water...sounds more like a Sandy Bridge chip your (running) thinking of

It will probably take a 5Ghz BD (with sub $100 WC) to outperform an SB at stock...

 

[Deleted member 217897]

There's a whole lot of "just pulled this out of may ass" going on in that reply!

hehe...I guess I can see where you get that from. I was basing the Ivy bridge points based on AT's recent article on Ivy Bridge.
http://www.anandtech.com/show/4830/intels-ivy-bridge-architecture-exposed

 

[daglesj]

People thought the world was flat at one time, so I don't put much stock into the past.

This is quite hilarious to me, the Intel corporate brainwashing here is out of control. I'll be the sanity since most* (you) people posting here have lost your minds to the big blue machine.

I'll be coming from an X3 @ 4.00GHz and going to a BD 8 Core that does 4.5 - 5.0 GHz on air and who knows what on water imo.

If you don't understand that it will rock, well then that's your problem.

And if all you have is made up expectations from incestuous Intel fans, and spouting off your negativity on AMD [H] forums about a CPU you haven't personally benched / used you just might be a fool.

AMD demo'd the chip OC'd it to 8.4+ if you can't believe reality, that the big red machine stomped every CPU in existence in OC well there's no helping you.

Chap, its not really down to that at all. It's that at 40 years of age with a wife, a mortgage and trying to keep ones head above water...I really dont care one way or another how AMD CPUs fare. Its not something I spend a lot of time worrying about. I'm interested, sure but its only silly old CPUs.

It's not really that important to me. I dont give a fig about Intel either. And if you think I care a jot about fansite benchmarks....oh boy.

I have a life chap.

 

[sirmo]

It will probably take a 5Ghz BD (with sub $100 WC) to outperform an SB at stock...

Even Phenom II could do that:

http://www.anandtech.com/bench/Product/362?vs=363

 

[teletran8]

It will probably take a 5Ghz BD (with sub $100 WC) to outperform an SB at stock...

AMD has already shown / proven just 1 of what I for-see as being many future points about why AMD have named their newer CPU model FX. The World Record OC.

The other being an 8 core CPU in today's market generation at a price FAR less than Intel's offering for 6 cores is really cutting edge imho.

What have you proven again? Oh yeah your a h8er.

 

[mzs_biteme]

Even Phenom II could do that:

http://www.anandtech.com/bench/Product/362?vs=363

You mean lose to a lowest end SB in all but one test, right?

 

[teletran8]

9-27-2011

Everyone would like to know the reason, but then again it's obviously a negative reason...yet no one (GF/AMD) is saying are they. Both of their buisnesses / corporations future is determined by product. Time will tell obviously. Only if yields are so bad they cannot make money AMD would be in trouble. Or simply AMD and/or GF has the time and is not into rushing, it could be as simple as that tho.

My opinion is that the new bulldozer "module" design leads to lower yields per wafer, or simply GF is slacking BIG time. They've obviously been able to yield a few though so things look fine to me...hehe.

^^
Told you guys the reason 2 days ago before that whack site Anandtech haha. Fun being right all the time hehe.

All they have to do @ GF is get their yields up, (just like we expierenced with Llano debut) but yeah we all now know BD will be in short supply for another quarter...damn that sux for them.

For whatever reason the yields are low, but since Llano is low ASWELL that makes me think it's not a Bulldozer flaw just a Global Foundries flaw in their setup / factory. Llano's are basically Phenom 2's with a mobile AMD 6 series GPU built in. But Llanos are in Stock on Newegg, so it's just a matter of time before we have some Bulldozers on Newegg and elsewhere.

There's a neat point in this article someone else had posted in another thread here and it stated that "In April, AMD said it changed its supply agreement with Globalfoundries to protect it against production delays. Under the revised agreement, AMD only pays for functioning chips delivered, rather than total production."

That's part if not all of the reason why Llano had a slower than wanted roll-out from AMD. GF's fab is running half ass these days, and AMD has more chips in demand than they used to globally.

So AMD has their shit together. They even called GF out on their "issues" and re=did their contract with them so they don't get burned financially. Only downside is that GF will most likely take their sweet a$$ time with it. But AMD dodged the bigger bullet IMO.

Now AMD just needs to keep us up on when those yields get fixed so we can really start getting excited. To bad I was the only one here thinking about YIELD difficulties while the rest of you were getting your panties in a bunch for months on end.

Just LOL!

 

[Zarathustra[H]]

Due in 2012?

I guess that means we'll see it sometime late 2016.

 

[teletran8]

Due in 2012?

I guess that means we'll see it sometime late 2016.

HAha let's hope Piledriver is not, or else Intel might go crazy (again) with their pricing.

GF needs to get their stuff together tho. They've been holding AMD back from the market somewhat.

 

[Michaelius]

HAha let's hope Piledriver is not, or else Intel might go crazy (again) with their pricing.

GF needs to get their stuff together tho. They've been holding AMD back from the market somewhat.

GF is AMD in all but name. They don't even have another client.

 

[Zarathustra[H]]

GF is AMD in all but name. They don't even have another client.

That is not correct.

Especially since the merger with Chartered Semiconductor last year.

"GlobalFoundries manufactures integrated circuits in high volume mostly for semiconductor companies such as AMD, Broadcom, Qualcomm, and STMicroelectronics."

AMD is by far their largest customer, but my guess would be that they account for only about 50% of their sales.

 

[Zarathustra[H]]

It will probably take a 5Ghz BD (with sub $100 WC) to outperform an SB at stock...

Posted elsewhere, but apparently it is needed here too...

Rumors have it that BD has no higher IPC than K10, and may even have taken a small step backwards in IPC due to the shared FPU.

If these rumors are true, this means that a BD would have to be clocked at somewhere from 5.2 to 5.6Ghz just to keep up with a stock clocked 3.4Ghz 2600k.

If you take into account that pretty much everyone with 2600k's are hitting 4.7 on air, BD would have to hit somewhere from 7.1 - 7.7Ghz on air to keep up... Assuming they even scale linearly up to the 5-7.5Ghz range...

This simply isn't happening.

Now, lets instead assume that they have done the impossible, and actually increased IPC over K10 by 20% in just one generation.

Even then, the numbers aren't that pretty:

BD would need to be clocked at 4.3 to 4.7Ghz, just to keep up with a stock 2600K

BD would have to be clocked at 5.9 - 6.5Ghz to keep up with an OC:ed 2600K @ 4.7Ghz.

I'm disappointed by this, as I am an AMD fan, but you just have to face reality at some point.

 

[teletran8]

Posted elsewhere, but apparently it is needed here too...

Rumors have it that BD has no higher IPC than K10, and may even have taken a small step backwards in IPC due to the shared FPU.

If these rumors are true, this means that a BD would have to be clocked at somewhere from 5.2 to 5.6Ghz just to keep up with a stock clocked 3.4Ghz 2600k.

If you take into account that pretty much everyone with 2600k's are hitting 4.7 on air, BD would have to hit somewhere from 7.1 - 7.7Ghz on air to keep up... Assuming they even scale linearly up to the 5-7.5Ghz range...

This simply isn't happening.

Now, lets instead assume that they have done the impossible, and actually increased IPC over K10 by 20% in just one generation.

Even then, the numbers aren't that pretty:

BD would need to be clocked at 4.3 to 4.7Ghz, just to keep up with a stock 2600K

BD would have to be clocked at 5.9 - 6.5Ghz to keep up with an OC:ed 2600K @ 4.7Ghz.

I'm disappointed by this, as I am an AMD fan, but you just have to face reality at some point.

Ah let's just wait and see what happens here, it will be released in some amount soon. Which means we will have performance numbers soon enough, wheather their benched in another country first doesn't matter to me, or even whether or not I can buy them first day I see them on Newegg doesn't matter to me.

The fact seems to be from AMD's own MARKETING DIVISION mouth that BD / Interlagos will be made available in the month of OCTOBER, (OCTO/ OCTOBER get it? ) and therefore we WILL see actual benchmark scores of it's performance.

Please STOP with all the guessing BS. It's annoying. We will have OFFICIAL numbers next month / October. Simple as that. No point looking into your CRYSTAL BALL of negativity BS again... haha.

If Novemeber rolls around and still nothing, PLEASE whip out that CRYSTAL BALL OF DOOM.

 

[drescherjm]

If Novemeber rolls around and still nothing, PLEASE whip out that CRYSTAL BALL OF DOOM.

LOL!

 

[sirmo]

Rumors have it that BD has no higher IPC than K10

You keep repeating this, over and over again. You never post sources, and I am forced to think the following:

How many times do I have to tell you that bulldozer has higher IPC than our current architecture?

Is somebody being paid by intel to continually post these statements?

source: http://www.xtremesystems.org/forums...nd-Bulldozer&p=4531177&viewfull=1#post4531177

Even Llano has 6% higher IPC than the current architecture, and no it's not just because of cache, which is another completely idiotic thing you keep polluting these forums with. Please stop trolling with your uninformed views for once, until you can either provide a shred of evidence that your post have any base on an actual believable piece of information.

 

[Zarathustra[H]]

You keep repeating this, over and over again. You never post sources, and I am forced to think the following:


source: http://www.xtremesystems.org/forums...nd-Bulldozer&p=4531177&viewfull=1#post4531177

Even Llano has 6% higher IPC than the current architecture, and no it's not just because of cache, which is another completely idiotic thing you keep polluting these forums with. Please stop trolling with your uninformed views for once, until you can either provide a shred of evidence that your post have any base on an actual believable piece of information.

If you read my post I stated I didn't know if it was true, but even if AMD pulled of a major upset and increased their IPC by 20% in one generation (which IMHO is impossible) the numbers still don't look good.

Regarding Llano, it's been repeated again and again that it is nothing but a K10 shrunk to 32nm with a GPU on die. Explain to me how that provides higher IPC? The basic architecture is identical. The only differences are the process size (which does nothing for IPC, but does a lot for temp/power/clock, and AMD don't seem to be taking advantage of the higher clock potential for some reason) and the rearrangement of the cache. (larger non-shared L2, but no L3).

 

[xorbe]

Ivy bridge is not any performance increase(atleast on the cpu side).

hehe...I guess I can see where you get that from. I was basing the Ivy bridge points based on AT's recent article on Ivy Bridge.
http://www.anandtech.com/show/4830/intels-ivy-bridge-architecture-exposed

I guess you didn't make it past page 1.

Reduction of HT penalty on statically split structures.
Prefetchers with less restriction.
FP/int divider faster / twice the throughput.
MOV op enhancement.
String op enhancements

 

[Tsumi]

If you read my post I stated I didn't know if it was true, but even if AMD pulled of a major upset and increased their IPC by 20% in one generation (which IMHO is impossible) the numbers still don't look good.

Regarding Llano, it's been repeated again and again that it is nothing but a K10 shrunk to 32nm with a GPU on die. Explain to me how that provides higher IPC? The basic architecture is identical. The only differences are the process size (which does nothing for IPC, but does a lot for temp/power/clock, and AMD don't seem to be taking advantage of the higher clock potential for some reason) and the rearrangement of the cache. (larger non-shared L2, but no L3).

Wasn't the jump from Core2Quad to Lynnfield/Nehalem something of that magnitude at least? That was a one generation jump...

Then again, I might be wrong, I didn't pay too much attention to Intel when they were releasing Lynnfield and Nehalem.

 

[sirmo]

Regarding Llano, it's been repeated again and again that it is nothing but a K10 shrunk to 32nm with a GPU on die. Explain to me how that provides higher IPC? The basic architecture is identical. The only differences are the process size (which does nothing for IPC, but does a lot for temp/power/clock, and AMD don't seem to be taking advantage of the higher clock potential for some reason) and the rearrangement of the cache. (larger non-shared L2, but no L3).

Thanks for proving my point:

http://images.anandtech.com/reviews/cpu/amd/llano/review/stars.jpg

Which is that you consistently keep trolling with your half-truths, in every single post.

 

[JMccovery]

If you read my post I stated I didn't know if it was true, but even if AMD pulled of a major upset and increased their IPC by 20% in one generation (which IMHO is impossible) the numbers still don't look good.

Regarding Llano, it's been repeated again and again that it is nothing but a K10 shrunk to 32nm with a GPU on die. Explain to me how that provides higher IPC? The basic architecture is identical. The only differences are the process size (which does nothing for IPC, but does a lot for temp/power/clock, and AMD don't seem to be taking advantage of the higher clock potential for some reason) and the rearrangement of the cache. (larger non-shared L2, but no L3).

They haven't increased the clocks on Llano to Deneb/Thuban levels or higher, because of the IGP, which probably is responsible for about half the power consumption and heat dissipation. Look at the A8-3850 and the A6-3650, both have a 100W TDP, but there is a 300MHz difference between them, similar to the Brazos chips, where even 'disabling' an entire core doesn't do much for power/TDP.

 

[yimingwuzere]

HAha let's hope Piledriver is not, or else Intel might go crazy (again) with their pricing.

GF needs to get their stuff together tho. They've been holding AMD back from the market somewhat.

Well we can wait till Q2 2012 and check out those with Trinity CPU benchmarks. Given that Trinity has two Bulldozer modules and AMD targets 50% higher total GFlops than Llano, I expect the cores to be a good deal better than the ones in Bulldozer.

On the other hand, Piledriver doesn't have the advantage Shanghai has over Barcelona (the smaller die size allows for higher clockspeeds at the same power).

 

[ChronoDetector]

AMD already starting to talk about the second generation and the first one hasn't even been released yet, but nice to hear about future upcoming hardware.

Anyway, I would prefer AMD to use a new entire new socket, they have used AMx for a long time now and its kinda holding them back. Would prefer to see more memory channels being implemented and new functions on an entirely new socket and chipset. Enough of this AMx backwards compatibility, they have done it to death, rather see them pull an Intel here.

 

[Zarathustra[H]]

Thanks for proving my point:

http://images.anandtech.com/reviews/cpu/amd/llano/review/stars.jpg

Which is that you consistently keep trolling with your half-truths, in every single post.

This is the only time I have seen a actual claim of higher IPC, and it is not explained in writing.

Every other review I have read on Llano speaks to the CPU portion simply being a die shrunk K10 core. You can't have it both ways. A simple die shrink without other modifications does not lead to any change in IPC at all.

Most reviews conclude that it must have something to do with the Cache. This may or may not be the case, as you can't test Llano with the same kind of cache as Phenom II, so there is no way to know for sure.

it is plausible though. Cache is no small deal. The L2 cache has doubled in size from 512kb per core, to 1mb per core. If there is an L2 cache miss, Phenom II had to go back to the L3 cache, which has less than half the bandwidth.

If there is an L2 cache miss on Llano - however - it has to go all the way back to system ram, as there is no L3 cache, so this makes it dreadfully slow.

So how should this show up in benchmarks?

Well, in benchmarks where most of the data computed fits in less than 512kb, they should be mostly equivalent.

In benchmarks where most of the data computed is larger than 512kb but smaller than 1mb, Llano ought to have a substantial performance benefit over Phenom II.

In benchmarks where most of the data computed exceeds 1mb, but is small enough to fit in the Phenom II's L3 cache, the Phenom II should outperform Llano, as system ram is comparatively slow compared to L3 cache, but with DDR3 at current speeds, the difference isn't as huge as it used to be, which is probably why they removed the L3 cache, as it wasn't doing them as much good anymore.

Most benchmarks have data all over the place. Sometimes data is already in a register and that clock it doesn't have to go to cache at all, other times it has to go to L1, or L2 cache, and sometimes out to l3 - if present - or RAM.

It is not inconceivable that on average the new larger L2 cache and removal of L3 cache would result in a noticeable performance increase. At the very least, the doubling of the ~30GB/s L2 cache ougt to improve performance more than the removal of the ~8GB/s L3 cache, especially since the L3 cache only has marginally higher bandwidth than RAM. (though the RAM has much higher latencies).

So long story short. All I am saying is that no one - except AMD R&D engineers can really say for sure where the performance increase is coming from. If it is - indeed - from IPC like you claim, then this is more than just a die shrink like it has been claimed over and over again. Simply changing the size of the process does not impact IPC, ever, unless accompanied by other design changes.

There is a solid case for the theory that the new Cache arrangement helps performance. as far as how much the cache helps vs how much any IPC tweaks may help it is impossible to test, so you'd have to know the explicit details of the engineering work on Llano, most of which is probably trade secrets.

 

[sirmo]

Doesn't change the fact that you consistently spew baseless claims. Like Llano is a simple die-shrink with no improvements whatsoever. Same with the claim that BD has a lower IPC thank K10.

Usually when people don't know anything about a subject they refrain from commenting about it.

 

[Digital Viper-X-]

Doesn't change the fact that you consistently spew baseless claims. Like Llano is a simple die-shrink which it wasn't. Same with the claim that BD has a lower IPC thank K10.

Usually when people don't know anything about a subject they refrain from commenting about it.

So not true, that statement looks better like this ..

Usually when people don't know anything about a subject they SHOULD refrain from commenting about it.

They can always give their opinions though

 

[JMccovery]

This is the only time I have seen a actual claim of higher IPC, and it is not explained in writing.

Every other review I have read on Llano speaks to the CPU portion simply being a die shrunk K10 core. You can't have it both ways. A simple die shrink without other modifications does not lead to any change in IPC at all.

Most reviews conclude that it must have something to do with the Cache. This may or may not be the case, as you can't test Llano with the same kind of cache as Phenom II, so there is no way to know for sure.

it is plausible though. Cache is no small deal. The L2 cache has doubled in size from 512kb per core, to 1mb per core. If there is an L2 cache miss, Phenom II had to go back to the L3 cache, which has less than half the bandwidth.

If there is an L2 cache miss on Llano - however - it has to go all the way back to system ram, as there is no L3 cache, so this makes it dreadfully slow.

So how should this show up in benchmarks?

Well, in benchmarks where most of the data computed fits in less than 512kb, they should be mostly equivalent.

In benchmarks where most of the data computed is larger than 512kb but smaller than 1mb, Llano ought to have a substantial performance benefit over Phenom II.

In benchmarks where most of the data computed exceeds 1mb, but is small enough to fit in the Phenom II's L3 cache, the Phenom II should outperform Llano, as system ram is comparatively slow compared to L3 cache, but with DDR3 at current speeds, the difference isn't as huge as it used to be, which is probably why they removed the L3 cache, as it wasn't doing them as much good anymore.

Most benchmarks have data all over the place. Sometimes data is already in a register and that clock it doesn't have to go to cache at all, other times it has to go to L1, or L2 cache, and sometimes out to l3 - if present - or RAM.

It is not inconceivable that on average the new larger L2 cache and removal of L3 cache would result in a noticeable performance increase. At the very least, the doubling of the ~30GB/s L2 cache ougt to improve performance more than the removal of the ~8GB/s L3 cache, especially since the L3 cache only has marginally higher bandwidth than RAM. (though the RAM has much higher latencies).

So long story short. All I am saying is that no one - except AMD R&D engineers can really say for sure where the performance increase is coming from. If it is - indeed - from IPC like you claim, then this is more than just a die shrink like it has been claimed over and over again. Simply changing the size of the process does not impact IPC, ever, unless accompanied by other design changes.

There is a solid case for the theory that the new Cache arrangement helps performance. as far as how much the cache helps vs how much any IPC tweaks may help it is impossible to test, so you'd have to know the explicit details of the engineering work on Llano, most of which is probably trade secrets.

Regor (Athlon II x2) has the same amount of L2 cache per core as Llano (1MB), but the 255 is a bit slower than the Phenom II 550: http://www.anandtech.com/bench/Product/121?vs=97

I don't think the extra cache alone is responsible, here is an AnandTech Bench comparison of the A8-3850 and Athlon II 635: http://www.anandtech.com/bench/Product/399?vs=122

AMD says that they improved the prefecther and increased the size of the reorder and load/store buffers, which COULD account for the minimal (~6%) IPC increase, but it is kinda hard to compare Llano to Deneb, since they have different cache setups. We will find out any differences once someone benches the Athlon II 631.

 

[Zarathustra[H]]

Doesn't change the fact that you consistently spew baseless claims. Like Llano is a simple die-shrink with no improvements whatsoever. Same with the claim that BD has a lower IPC thank K10.

Usually when people don't know anything about a subject they refrain from commenting about it.

So you are saying I need to cite references for everything I post? I don't have time for such nonsense. I know what I have read. This is not a scientific paper, it's a forum.

 

[Zarathustra[H]]

Doesn't change the fact that you consistently spew baseless claims. Like Llano is a simple die-shrink with no improvements whatsoever. Same with the claim that BD has a lower IPC thank K10.

Usually when people don't know anything about a subject they refrain from commenting about it.

I have never claimed that Bulldozer has lower IPC than K10.

I have - however - stated that I have heard rumors to this effect. Rumors are exactly that, unverifiable. Whenever I - or anyone else - type the word "rumor" it means that it should be taken with a grain of salt.

This is also why I followed up by stating that even if they made a 20% improvement over k10 in one generation (which would be unheard of) the numbers don't look great.

You need to just chill the fuck out.

 

[sirmo]

I have never claimed that Bulldozer has lower IPC than K10.

I have - however - stated that I have heard rumors to this effect. Rumors are exactly that, unverifiable. Whenever I - or anyone else - type the word "rumor" it means that it should be taken with a grain of salt.

This is also why I followed up by stating that even if they made a 20% improvement over k10 in one generation (which would be unheard of) the numbers don't look great.

You need to just chill the fuck out.

I think you're starting those rumors though. Even rumors have an origin. And as far as your posts go you're the origin of these rumors.

Same way you make posts in which you claim, with certainty, that Llano is nothing but a die shrink with no tweaks whatsoever, even though the opposite it's well established.

You are spreading misinformation with your every post and I am tired of it. Uninformed opinions, rumors and statements, is not why we all come to [H] forums.

 

[Zarathustra[H]]

I think you're starting those rumors though.

Uh, no. I remember seeing a few headlines on the internets in the last couple of months drawing this conclusion. When I get out of work I'll have to do some googling and look them up for you

You are spreading misinformation with your every post and I am tired of it. Uninformed opinions, rumors and statements, is not why we all come to [H] forums.

Rumors are all we have to go by right now, since AMD has been less than forthcoming with this launch. There is nothing wrong with speculating based on rumors, as long as you state that is what you are doing.

I never claimed to be a chip or even software engineer, but I do have a layman's understanding of the subject, and I will post my opinions based on th egeneral impressions I get and on the data I have seen.

If you have a problem with that, I suggest you not only leave this forum, but leave all forums on the internet. That's how they work. Deal or QQ.

As has been stated earlier, we will know how it truly performs when it comes out. Until then a little speculation is only healthy. I'm buying one either way, as it most certainly will be an overall improvement over my Thuban, but anyone who expects it to be an Intel killer, or even able to match Sandy Bridge in anything but the likes of mulithreaded Cinebench is sadly very delusional at this point.